Data extraction using different trained models

Abstract

Systems and methods reception of an object for entity extraction, identification of an extraction schema instance associated with the object, determination of a first extraction field and a first model associated with the first extraction field based on the extraction schema instance, generation of an input payload according to an input format of the first model, reception of a first value of the first extraction field output by the first model, determination of a second extraction field and a second model associated with the second extraction field, input of the object to the second model to output a second value of the second extraction field, and reception of the second value.

What is claimed is: 
     
         1 . A system comprising:
 a memory storing processor-executable program code; and   a processing unit to execute the processor-executable program code to cause the system to:   receive an object on which to perform entity extraction;   from a plurality of extraction schema instances, identify an extraction schema instance associated with the object;   determine a first extraction field and a first model associated with the first extraction field based on the extraction schema instance;   generate an input payload according to an input format of the first model;   transmit the input payload to the first model;   receive a first value of the first extraction field output by the first model;   determine a second extraction field and a second model associated with the second extraction field based on the extraction schema instance;   input the object to the second model to output a second value of the second extraction field; and   receive the second value.   
     
     
         2 . A system according to  claim 1 , wherein the first model is a large language model, and the input payload includes a prompt and the object. 
     
     
         3 . A system according to  claim 2 , wherein the second model is a pre-trained model. 
     
     
         4 . A system according to  claim 1 , the processing unit to execute the processor-executable program code to cause the system to:
 determine a confidence score associated with the second value;   determine a threshold associated with the second extraction field based on the instance;   determine if the confidence score is greater than the threshold; and   if the confidence score is not greater than the threshold:   determine a third model associated with the second extraction field based on the extraction schema instance;   generate a second input payload according to an input format of the third model;   transmit the second input payload to the third model;   receive a third value of the second extraction field output by the third model; and   return the first value and the third value.   
     
     
         5 . A system according to  claim 4 , the processing unit to execute the processor-executable program code to cause the system to:
 receive a second object on which to perform entity extraction;   from a plurality of extraction schema instances, identify the extraction schema instance as associated with the second object;   determine the first extraction field and the first model associated with the first extraction field based on the extraction schema instance;   generate a third input payload according to the input format of the first model;   transmit the third input payload to the first model;   receive a fourth value of the first extraction field output by the first model;   determine the second extraction field and the second model associated with the second extraction field based on the extraction schema instance;   input the second object to the second model to output a fifth value of the second extraction field;   determine a second confidence score associated with the fifth value;   determine that the second confidence score is greater than the threshold; and   based on the determination that the second confidence score is greater than the threshold, return the fourth value and the fifth value.   
     
     
         6 . A system according to  claim 1 , the processing unit to execute the processor-executable program code to cause the system to:
 receive a second object on which to perform entity extraction;   from a plurality of extraction schema instances, identify a second extraction schema instance associated with the second object;   determine the first extraction field and the first model associated with the first extraction field based on the second extraction schema instance;   generate a second input payload according to the input format of the first model;   transmit the second input payload to the first model;   receive a third value of the first extraction field output by the first model;   determine a third extraction field and a third model associated with the third extraction field based on the second extraction schema instance;   generate a third input payload according to the input format of the third model;   transmit the third input payload to the third model; and   receive a fourth value of the third extraction field output by the third model.   
     
     
         7 . A system according to  claim 1 , the processing unit to execute the processor-executable program code to cause the system to:
 receive a second object on which to perform entity extraction;   from a plurality of extraction schema instances, identify a second extraction schema instance associated with the second object;   determine the first extraction field and the first model associated with the first extraction field based on the second extraction schema instance;   generate a second input payload according to the input format of the first model;   transmit the second input payload to the first model;   receive a third value of the first extraction field output by the first model;   determine the second extraction field and a third model associated with the second extraction field based on the second extraction schema instance;   generate a third input payload according to the input format of the third model;   transmit the third input payload to the third model; and   receive a fourth value of the second extraction field output by the third model.   
     
     
         8 . A method comprising:
 receiving a document;   from a plurality of extraction schema instances, identifying an extraction schema instance based on a type of the document;   determining a first extraction field and a first model associated with the first extraction field based on the extraction schema instance;   generating an input payload including text of the document and according to an input format of the first model;   transmitting the input payload to the first model;   receiving a first value of the first extraction field output by the first model;   determining a second extraction field and a second model associated with the second extraction field based on the extraction schema instance;   inputting the text of the document to the second model to output a second value of the second extraction field; and   receiving the second value.   
     
     
         9 . A method according to  claim 8 , wherein the first model is a large language model, and the input payload includes a prompt and the text of the document. 
     
     
         10 . A method according to  claim 9 , wherein the second model is a pre-trained model. 
     
     
         11 . A method according to  claim 8 , further comprising:
 determining a confidence score associated with the second value;   determining a threshold associated with the second extraction field based on the instance;   determining if the confidence score is greater than the threshold; and   if the confidence score is not greater than the threshold:   determining a third model associated with the second extraction field based on the extraction schema instance;   generating a second input payload including the text and according to an input format of the third model;   transmitting the second input payload to the third model;   receiving a third value of the second extraction field output by the third model; and   returning the first value and the third value.   
     
     
         12 . A method according to  claim 11 , further comprising:
 receiving a second document;   generating a third input payload including text of the second document and according to the input format of the first model;   transmitting the third input payload to the first model;   receiving a fourth value of the first extraction field output by the first model;   inputting the text of the second document to the second model to output a fifth value of the second extraction field;   determining a second confidence score associated with the fifth value;   determining that the second confidence score is greater than the threshold; and   based on the determination that the second confidence score is greater than the threshold, returning the fourth value and the fifth value.   
     
     
         13 . A method according to  claim 8 , further comprising:
 receiving a second document;   identifying a second extraction schema instance associated with the second document;   determining the first extraction field and the first model associated with the first extraction field based on the second extraction schema instance;   generating a second input payload according to the input format of the first model;   transmitting the second input payload to the first model;   receiving a third value of the first extraction field output by the first model;   determining a third extraction field and a third model associated with the third extraction field based on the second extraction schema instance;   generating a third input payload according to the input format of the third model;   transmitting the third input payload to the third model; and   receiving a fourth value of the third extraction field output by the third model.   
     
     
         14 . A method according to  claim 8 , further comprising:
 receiving a second document;   identifying a second extraction schema instance associated with the second document;   determining the first extraction field and the first model associated with the first extraction field based on the second extraction schema instance;   generating a second input payload according to the input format of the first model;   transmitting the second input payload to the first model;   receiving a third value of the first extraction field output by the first model;   determining the second extraction field and a third model associated with the second extraction field based on the second extraction schema instance;   generating a third input payload according to the input format of the third model;   transmitting the third input payload to the third model; and   receiving a fourth value of the second extraction field output by the third model.   
     
     
         15 . A non-transitory medium storing processor-executable program code executable by a processing unit of a computing system to cause the computing system to:
 receive an object on which to perform entity extraction;   from a plurality of extraction schema instances, identify an extraction schema instance associated with the object;   determine a first extraction field and a first model associated with the first extraction field based on the extraction schema instance;   generate an input payload according to an input format of the first model;   transmit the input payload to the first model;   receive a first value of the first extraction field output by the first model;   determine a second extraction field and a second model associated with the second extraction field based on the extraction schema instance;   input the object to the second model to output a second value of the second extraction field; and   receive the second value.   
     
     
         16 . A medium according to  claim 15 , wherein the first model is a large language model, and the input payload includes a prompt and the object. 
     
     
         17 . A medium according to  claim 15 , the processor-executable program code executable by a processing unit of a computing system to cause the computing system to:
 determine a confidence score associated with the second value;   determine a threshold associated with the second extraction field based on the instance;   determine if the confidence score is greater than the threshold; and   if the confidence score is not greater than the threshold:   determine a third model associated with the second extraction field based on the extraction schema instance;   generate a second input payload according to an input format of the third model;   transmit the second input payload to the third model;   receive a third value of the second extraction field output by the third model; and   return the first value and the third value.   
     
     
         18 . A medium according to  claim 17 , the processor-executable program code executable by a processing unit of a computing system to cause the computing system to:
 receive a second object on which to perform entity extraction;   from a plurality of extraction schema instances, identify the extraction schema instance as associated with the second object;   determine the first extraction field and the first model associated with the first extraction field based on the extraction schema instance;   generate a third input payload according to the input format of the first model;   transmit the third input payload to the first model;   receive a fourth value of the first extraction field output by the first model;   determine the second extraction field and the second model associated with the second extraction field based on the extraction schema instance;   input the second object to the second model to output a fifth value of the second extraction field;   determine a second confidence score associated with the fifth value;   determine that the second confidence score is greater than the threshold; and   based on the determination that the second confidence score is greater than the threshold, return the fourth value and the fifth value.   
     
     
         19 . A medium according to  claim 17 , the processor-executable program code executable by a processing unit of a computing system to cause the computing system to:
 receive a second object on which to perform entity extraction;   from a plurality of extraction schema instances, identify a second extraction schema instance associated with the second object;   determine the first extraction field and the first model associated with the first extraction field based on the second extraction schema instance;   generate a second input payload according to the input format of the first model;   transmit the second input payload to the first model;   receive a third value of the first extraction field output by the first model;   determine a third extraction field and a third model associated with the third extraction field based on the second extraction schema instance;   generate a third input payload according to the input format of the third model;   transmit the third input payload to the third model; and   receive a fourth value of the third extraction field output by the third model.   
     
     
         20 . A medium according to  claim 17 , the processor-executable program code executable by a processing unit of a computing system to cause the computing system to:
 receive a second object on which to perform entity extraction;   from a plurality of extraction schema instances, identify a second extraction schema instance associated with the second object;   determine the first extraction field and the first model associated with the first extraction field based on the second extraction schema instance;   generate a second input payload according to the input format of the first model;   transmit the second input payload to the first model;   receive a third value of the first extraction field output by the first model;   determine the second extraction field and a third model associated with the second extraction field based on the second extraction schema instance;   generate a third input payload according to the input format of the third model;   transmit the third input payload to the third model; and   receive a fourth value of the second extraction field output by the third model.